Our June SF Metrics Meetup featured Rob Miller from Mozilla and Noah Kantrowitz. For those of you who couldn't make it, here are the recordings:
Stream Processing Made Heka Easy
Rob Miller, Mozilla
Do you have servers? Are they running services? Are all of those servers and services producing data? Do you ever feel overwhelmed by all of that data, and wish that it were easier to parse, crunch, process, aggregate, watch, ship, and generate notifications based on that data? In this talk Rob Miller shows how Mozilla's Cloud Services team has been using their stream processing tool Heka to tame the firehose.
Is There An Echo In Here?
For decades, audio engineers have been finding new and better ways to improve the quality of their signals. From the basics of high-pass and low-pass filters, to modern anti-feedback and echo cancellation algorithms, there is a huge body of work both academic and practical. Operations monitoring and metrics tools have also advanced rapidly in the last few years, leading to vastly improved alerting and general awareness of system health.
This is not without a downside, however. As anyone that has been on the receiving end of Nagios' flap detection or a graphite-induced pager storm can attest, the signal to noise ratio (itself an audio engineering term) has become a problem for many organizations. At heart a stream of audio data and a stream of metrics data are quite similar. This talk explores some basic algorithms from Digital Signal Processing you can use to level up your monitoring systems.